Affordable and fair?Analyzing transit fare purchases, service quality, and affordability and their implications for social equity

A supervised research project byDavid Verbich

Supervised byAhmed El-Geneidy

Submitted toSchool of Urban PlanningMcGill University

April 2016

Affordable and fair? Analyzing transit fare purchases, service quality, and

affordability and their implications for social equity

By

David Verbich

Supervised by Ahmed El-Geneidy

School of Urban Planning

McGill University

Montréal, Québec, Canada

April 2016

Supervised Research Project Report

Submitted in partial fulfillment of the requirements of the degree of Master of Urban Planning

"Every city that acts like a city has transit."

—Frederick P. Salvucci

i

TABLE OF CONTENTS

ABSTRACT iii

RÉSUMÉ v

ACKNOWLEDGMENTS vii

CHAPTER 1. INTRODUCTION 1

CHAPTER 2. PUBLIC TRANSIT FARE STRUCTURE AND INCOME VULNERABILITY IN

MONTREAL, CANADA 4

2.1 INTRODUCTION 4 2.2 LITERATURE REVIEW 5 2.3 DATA AND METHODOLOGY 7

2.3.1 Montreal background and context 7

2.3.2 Data preparation 9

2.3.3 Income and spatial variables 11

2.3.4 Statistical modelling 11

2.4 RESULTS 12

2.4.1 Transit fare purchase models 14

2.5 DISCUSSION AND CONCLUSION 18

CHAPTER 3. HOW AFFORDABLE IS QUALITY PUBLIC TRANSPORTATION FOR MINIMUM

WAGE EARNERS IN NORTH AMERICA? 22

3.1 INTRODUCTION 22

3.2 LITERATURE REVIEW 23 3.3 METHODOLOGY 26 3.4 RESULTS AND DISCUSSION 31

3.4.1 Rider perspective 31

3.4.2 Society and transit agency perspectives 36

3.4.3 Overall service quality 37

CHAPTER 4. AFTERWORD AND CONCLUSIONS 44

REFERENCE LIST 46

ii

LIST OF FIGURES AND TABLES

Figure 2.1. Context map of the Island of Montreal. 8 Figure 2.2. Data preparation. 10

Figure 2.3. Income and fare purchase deciles. 14 Figure 3.1. Balancing affordability and transit service quality from multiple perspectives. 26

Figure 3.2. Accessibility, headway, and comfort indicators plotted against a measure of affordability. Note that only headway is normalized since this indicator is composed of two measures (headways at peak and off-peak hours). 35

Figure 3.3. Rider, society, and agency indicators plotted against a measure of affordability. 39 Figure 3.4. Service quality indicator plotted against a measure of affordability. 41

Table 2.1. Description of Variables and Summary Statistics. 13

Table 2.2. Model 1 – Monthly fare purchase model. 15 Table 2.3. Model 2 – Weekly fare purchase model. 16

Table 2.4. Model 3 – Three or more recurrent weekly fare purchase model. 18 Table 3.1. Cities and transit agencies. 27 Table 3.2. Transit rider indicators. 32

Table 3.3. Transit agencies, fares, minimum wages, and affordability. 33 Table 3.4. Ridership per capita and society indicator. 36

Table 3.5. Farebox recovery ratio (FRR) and agency indicator. 37 Table 3.6. Final ranking, service quality indicator and affordability 40

iii

ABSTRACT

Public transport agencies have the difficult task of providing service that meets the desires and needs of

the residents of a city while remaining financially viable. Unfortunately, in North America, funding

cutbacks often precipitate either service adjustments or fare increases, or often both. As a result, transit

may become a less attractive for choice riders, but may become altogether unaffordable and

inconvenient for captive riders who are typically marginalized members of society and have no choice

but to rely on public transport for travel. In this research project, I investigate how median income of a

neighbourhood influences transit fare purchases, as well as the affordability and service quality supplied

by transit agencies in the fourteen largest North American cities.

First, I studied whether the type of transit fare purchased, either a monthly fare or a weekly fare,

depends on income. I discovered that in Montreal, lower incomes increase the purchase of multiple

weekly fares. This finding suggests that individuals residing in low-income neighbourhoods may

purchase weekly fares as a substitute for monthly fares since monthly fares require greater upfront

costs. Moreover, tied to the idea of transit affordability is whether a transit agency can supply quality

service at an affordable price. To investigate this issue, I used publicly-available data to assess the

performance of transit agencies based on a parsimonious set of indicators aimed at capturing the

perspective of transit riders, society as a whole, and transit agencies themselves. Overall, transit

agencies with the best quality are relatively expensive for minimum wage earners, and worse still, some

agencies provide poor quality service that is also expensive. Nevertheless, the SFMTA in San Francisco

and the STM in Montreal are two agencies that are the most affordable that also provide good service

to their riders.

My findings indicate that low-income riders may be overburdened by transit fares and call for

reduced fares for low-income earners or the adoption of a new fare policy, such a fare capping.

Moreover, my research suggests that North American transit agencies have difficulty providing quality

service that is also affordable for minimum wage earners. Taken together, my research adds to the

growing body of evidence demonstrating the difficulty that disadvantaged residents face when

iv

financially accessing public transport. From a social equity perspective, in order to provide financially

viable transit service that meets the needs of all, transit fares should be harmonized with the ability pay.

v

RÉSUMÉ

Les agences métropolitaines de transport ont un mandat difficile de fournir un service qui répond aux

désirs et aux besoins des résidents, tout en restant financièrement viables. Malheureusement, les

agences se voient souvent dans l’obligation de réduire leur service ou d’augmenter leur prix en raison

de coupures dans le financement du transport en commun. Ces changements peuvent rendre le

service moins attirant pour l’ensemble des usagers. De plus, le transport en commun peut devenir non

abordable pour les usagers vulnérables qui dépendent du transport en commun pour leurs

déplacements quotidiens. L’objectif de cette recherche est donc d’explorer dans quelle mesure le

revenu médian d’un quartier influence l’achat de billets de transport en commun, et d’analyser la

qualité, et l’accessibilité financière, du service fournie par les agences de transport en commun dans les

14 plus grands villes américaines.

En premier lieu, j’ai étudié si le type de billet acheté (mensuel ou hebdomadaire) était influencé

par le revenu. Les résultats indiquent que, à Montréal, un revenu plus faible augmente l’achat de

passes hebdomadaires multiples, comparativement à l’achat d’une passe mensuelle. Ces résultats

suggèrent que les individus résidant dans des quartiers dont le revenu médian est plus faible pourraient

acheter de multiples passes hebdomadaires en remplacement d’une passe mensuelle, étant donné

qu’une passe mensuelle requiert un plus grand débours au début du mois. De plus, une autre question

liée à l’accessibilité financière réside dans la capacité des agences à fournir un service de qualité, à un

prix abordable. Pour explorer cet enjeu, j’ai utilisé des données publiques me permettant d’analyser la

performance des agences de transport, en fonction d’une série d’indicateurs visant à cerner la

perspective des usagers, de l’ensemble de la société, et des agences de transport. De façon générale,

les agences de transport fournissant les services de plus haute qualité s’avèrent relativement

dispendieuses pour les individus à salaire minimum. Par ailleurs, certaines agences fournissant un

service de faible qualité sont elles aussi dispendieuses. Néanmoins, la SFMTA à San Francisco et la

STM à Montréal sont les deux agences les plus abordables qui fournissent aussi un bon service à leurs

usagers.

vi

Les résultats de ma recherche indiquent que les usagers à faible revenu peuvent porter un

fardeau important en termes de coût du transport et mettent en perspective la nécessité d’implanter

des tarifs réduits pour les individus à faible revenu, ou l’adoption d’une nouvelle politique tarifaire, telle

qu’un plafond tarifaire. De plus, les résultats suggèrent que les agences de transport nord-américaines

ont de la difficulté à fournir un service de qualité qui soit abordable pour les moins nantis.

Dans l’ensemble, ma recherche s’ajoute aux preuves grandissantes démontrant les difficultés

auxquelles font face les résidents vulnérables en matière d’accessibilité financière. Dans une

perspective d’équité sociale, les tarifs de transport en commun devraient être harmonisés avec la

capacité de payer de chacun des usagers, afin de fournir un système de transport en commun

financièrement viable qui rencontre les besoins de tous les résidents d’une région urbaine.

vii

ACKNOWLEDGMENTS

I would first like to thank Ahmed for being an excellent supervisor who taught me so much about planning and about myself, constantly pushing me to think about my results and fostering creativity. Thank you for always being available to discuss ideas and anything else planning and non-planning related. Thank you Madhav for being a second reader and providing valuable advice and input on this SRP and for the thought-provoking discussions. I wish to thank everyone at TRAM for the advice related to projects but as well as providing support and laughs through tough and fun times, particularly at TRB. So thanks Dea, Gen, Charis, Emily and Ehab, as well as previous students Alex, Myriam, and Scott. Special thanks goes to Gen for helping me translate mon résumé (on a tight deadline) and for help with accessibility measures. In addition, I would like to thank Marc-André Gatien from the STM for the data used in Chapter 2, and Rania Wasfi for advice and help with multilevel modelling. I wish to also thank Brendon Hemily for advice and fare data used in Chapter 3, Lauren Rudko from CUTA for data on Canadian transit agencies, and Andrew Owen and David Levinson for accessibility measures for American cities. I wish to thank SSHRC for funding support and Gladys Chan and Anand Sood for excellent administrative help throughout the Master's. I want to thank Zeina for critical discussion and support during the SRP process. I want to thank my sisters for all their support and inspiration. Finally, I wish to thank my parents, who are my biggest supporters and who did not disown a son who finished a PhD in neuroscience only to do a Master's in urban planning.

1

CHAPTER 1. INTRODUCTION

Public transit is increasing in popularity and use. For example, last October saw the most trips taken on

the New York City subway for a weekday since the 1940s (MTA, 2015). Indeed, after reaching a nadir

during the 1970s and 1980s, reinvestments in public transit in the United States and Canada in the

1990s have helped grow ridership (Seaman, de Cerreno, & English-Young, 2004). The desire to

increase public transit use is consistent with the focus of making cities more sustainable and liveable.

Who in a city has access to public transit is an important concern, particularly since

transportation projects are mainly paid for with public funds (Manaugh & El-Geneidy, 2012; Martens,

Golub, & Robinson, 2012). As such, an equitable distribution of transportation infrastructure and the

benefits accrued from this infrastructure are goals in many city and transport agency planning

documents, although concretizing these goals is difficult and sometimes elusive (Manaugh, Badami, &

El-Geneidy, 2015). Contemporary research in social equity and public transit tends to focus on

accessibility and whether disadvantaged populations are able to access desired destinations, such as

jobs and social opportunities (Delbosc & Currie, 2011; Foth, Manaugh, & El-Geneidy, 2013; Geurs &

van Wee, 2004; Litman, 2002). In addition to physically accessing transit, city residents must also be

able to financially afford public transit.

Transit fares may pose a barrier to some people, particularly low-income people, such as single

parents, elderly people on a pension, and students (Taylor & Norton, 2009). While discounted transit

fares are typically offered to students and the elderly, fewer transit agencies offer discounts for low-

income populations (Taylor & Jones, 2012). Furthermore, transit agencies attempt to achieve financial

solvency which typically depends on transit fares, while providing an acceptable level of service at an

affordable price (Badami & Haider, 2007). While transit agencies can raise fares to cover revenue

generation related to investments or capital programs (Aggarwala, 2012), in the United States,

proposed fare increases must undergo scrutiny to ensure compliance with civil rights legislation,

2

essentially safeguarding minority and low-income groups from burdensome transit fares (Grengs, 2002;

METRO, 2016).

The overarching theme of this research project is the financial affordability of public transit. In

particular, this work was inspired by a finding that entries into New York City subway stations located in

many low-income census tracts are predominant based on the use of weekly passes, while entries in

high-income census tracts are mostly from monthly passes (Hickey, Lu, & Reddy, 2010). This

observation is germane given that weekly passes, while cheaper to purchase upfront, cost more if

purchased repeatedly over the course of a month as a substitute for a monthly pass—buying in bulk is

cheaper (Schuerman, 2015). In particular, low-income earners may be unable to put down the large

sum required for a monthly pass and opt instead to purchase weekly passes (Provost, 2015;

Schuerman, 2015). Since they end paying more over the long-term than their high-income

counterparts, this situation overburdens already vulnerable people. Therefore, in Chapter 2 of this

project, I investigated how transit fare purchases on the Island of Montreal may be influenced by

median income. I found that at the neighbourhood level, lower incomes predict greater multiple weekly

fare purchases than in neighbourhoods with higher incomes. These results suggest that in Montreal,

residents of marginalized neighbourhoods may purchase weekly fares as substitutes for monthly fares,

and as a result, spend more money than residents in wealthier neighbourhoods on transit passes.

In response to these findings, a spokesperson from the Montreal transit corporation countered

that among North American agencies, Montreal has some of the most affordable fares (Provost, 2015).

This prompted a number of questions: If Montreal’s transit is considered affordable, does the transit

agency provide good quality service? If other agencies have more expensive fares, do they provide

better service? Are there agencies that provide as good or better service more affordably? And even if

the cost of fares is lower in Montreal than in other cities, how affordable is it for minimum wage earners

compared to other cities? To begin answering some of these questions, in Chapter 3, I present a critical

analysis of the affordability and the quality of service offered by transit agencies in large North American

cities. Using publicly-reported transit data, I created indicators based on features desired by transit

3

riders: accessibility to jobs, service frequency, and comfort. Using the hourly minimum wage, I also

calculated the number of hours needed to purchase a monthly transit fare as a measure of affordability

for low-income users and compared the different quality indicators with affordability. In addition, I also

considered the perspective of financial viability of transit agencies and the benefits provided by public

transport to society by having a system that is well-used. Overall, by combining different perspectives, I

discovered that the best performing agencies are typically expensive for minimum wage earners, except

for transit in Montreal and San Francisco, where their respective transit agencies seem to provide

quality and affordable service. With this methodology, I provide a framework for researchers and

practitioners wishing to assess and compare the quality and affordability of public transport systems.

4

CHAPTER 2. PUBLIC TRANSIT FARE STRUCTURE AND INCOME VULNERABILITY IN

MONTREAL, CANADA1

2.1 INTRODUCTION

A public transit network capable of moving many residents to diverse locations is an integral

characteristic of any city. A large focus of public transit research deals with scheduling and operations

or travel behavior and mode share, but fewer studies have investigated fare structures and purchasing.

Moreover, from a social equity perspective, a growing body of literature has asked whether transit

benefits and projects are distributed equitably in a region (Delbosc & Currie, 2011; Foth et al., 2013;

Manaugh et al., 2015; Martens, 2012), but few studies have addressed the equity impacts of fare

structures. In fact, transit fares may form an important barrier for the working poor (Stolper & Rankin,

2016). One recent example of research on fare structures was a study in New York City, where

researchers discovered that entries at subway stations in low-income census tracts were predominately

from seven-day fare cards, while entries at stations in higher income census tracts were largely from

monthly fare cards (Hickey et al., 2010). In New York City—as in many other cities—purchasing multiple

weekly fares as a substitute for a monthly fare costs more in the long run since buying in bulk is

cheaper. However, low-income earners may be unable to spend a large amount at one time for a

monthly fare, opting instead to buy weekly fares that are cheaper in the short-term but costlier in the

long-term (Schuerman, 2015; Stolper & Rankin, 2016).

In this paper, we investigate how purchases of different transit fare types may relate to income,

with implications for social equity. We studied purchases of transit fares using OPUS card transaction

records, the transit fare smartcard of the transit agency of the Island of Montreal, Canada, the Société

de transport de Montréal (STM) (STM, 2015a). To shed light on the spatial and socioeconomic factors

that underlie different types of fare purchases, we model the number of total monthly fares purchased,

1 This chapter is currently under review as a manuscript at Transportation Research Part A: Policy and Practice.

5

total weekly fares purchased, and the number of riders who purchased three or more weekly fares

during September 2014. We hypothesize that some low-income earners may be unable to purchase

monthly passes because of the high upfront cost, and buy weekly passes as a substitute; while a

weekly pass is less expensive upfront, buying three or more weekly passes over the course of the

month is costlier (per ride). Based on our hypothesis, we expect to find concentrations of weekly fare

transactions predominately in low-income neighbourhoods, and we also predict that income will be a

significant explanatory variable of three or more weekly fare sales that we hereafter refer to

interchangeably as recurring, repeated, or multiple weekly fare sales.

The paper begins with a review of the relevant literature on transport equity and transit fares.

Next, we provide background on the Montreal context and transit fare structure of the STM, Montreal’s

main transit provider. Third, we describe our dataset, methods, and modelling approach. Fourth, we

explain our models and findings. Lastly, we discuss our findings and potential policy relevance.

2.2 LITERATURE REVIEW

An emerging body of literature has brought social equity concerns to the forefront of many disciplines,

including land use and transportation planning (Geurs, Boon, & Van Wee, 2009; Martens, 2012).

Nevertheless, in many instances, what equity entails is murky including equity in transportation planning

(Manaugh et al., 2015) and the definition of equity itself (Taylor & Norton, 2009). Generally, equity in

reference to planning, as espoused by Krumholz and Forester (1990), means allocating more resources

to those who have the least. This idea would propose ensuring that socially vulnerable populations, who

are typically bounded to transit (APTA, 2007), have a larger share of transportation-related benefits,

such as accessibility to jobs by transit (Foth et al., 2013) or access to transit itself (Delbosc & Currie,

2011; Welch, 2013), than individuals with more transport options, like a private vehicle. With regard to

transit fares, an equitable scheme consistent with Krumholz and Forester's (1990) thesis would be one

where low-income individuals pay less, while individuals who can afford to pay more will pay a larger

fare (Aggarwala, 2012).

6

Many different fare schemes are used by transit agencies with varying impacts on social equity.

In North American cities, many transit agencies charge flat fares for city bus and subway services

(Cervero, 1981), regardless of distance travelled or mode. Since low-income residents tend to live in

central city neighbourhoods well-served by transit (Glaeser, Kahn, & Rappaport, 2008), these residents

likely travel shorter distances and thus are poorly served by flat fares. On the other hand, commuter rail

tends to be zonal in fare structure and is mostly used by affluent white-collar workers (Taylor & Morris,

2015). These schemes also apply to Montreal, where one flat fare is used for city buses and metros,

but commuter rail fares are zonal. Recent work using spatial modelling to examine a switch from flat to

distance-based fares found that this switch is advantageous for minority and low-income riders in

Wasatch Front, Utah (Farber, Bartholomew, Li, Páez, & Nurul Habib, 2014). In addition to considering

fares based on distance or time of day, different fares types, such as monthly unlimited or weekly

unlimited passes are also used by many agencies. The price-per-ride of a monthly pass is usually less

expensive than weekly or daily single fares.

With varying fare schemes, demand for transit can be managed. For example, longer trips in

Beijing are most affected by fare increases (Wang, Li, & Chen, 2015), while in New York City, estimated

ridership loss is less for increases in unlimited monthly fares compared to weekly fare increases (Hickey,

2005), implying that sensitivity of ridership depends on fare structure and type. By evaluating several

scenarios of Alameda-Contra Costa Transit, California, Nuworsoo, Golub, and Deakin (2009) found that

vulnerable riders, including low-income riders, youth, and minorities would be penalized more by

scenarios that would increase flat fares per ride due to more transfers and more trips compared to

affluent riders who could afford to purchase a monthly pass. In contrast to agencies with flat fares or

costlier fares for fewer rides, in Washington, D.C., distance and peak trips determine transit fares, and

per ride, weekly fares cost the same as monthly fares (WMATA, 2015). Washington’s approach to

charging lower fares during off-peak hours could address some equity concerns (El-Geneidy et al.,

2015; Taylor & Norton, 2009). To specifically address vulnerable populations who depend on transit,

7

transit agencies in San Francisco, Seattle, and Calgary, for example, have programs for subsidized

transit passes for low-income riders (SFMTA, n.d.; King County Transit, 2015; Calgary Transit, 2015).

Nevertheless, few studies have considered the factors underlying transit fare purchases. By

estimating travel demand in Athens, Greece, researchers uncovered that the fare type can influence

ridership, and this varies by mode as well as fare price (Gkritza, Karlaftis, & Mannering, 2011). While that

study used some sociodemographic indicators, namely income and immigrant status, to determine

their association with fare purchases, detailed demographics were lacking. This is likely due to the

inability to link riders and demographics directly. Given the scarce research on income and transit fare

purchasing with implications for social equity, a closer examination of how income can influence transit

fare purchases is warranted.

2.3 DATA AND METHODOLOGY

2.3.1 Montreal background and context

Montreal is the second largest metropolitan region in Canada, at roughly 4 million inhabitants, and is

located in the south of the province of Quebec. The City of Montreal, along with on-island

municipalities, is located on Montreal Island in the Saint Lawrence River, and is served by the STM, the

local transit agency operating the metro and buses on the Island. The regional transit authority runs

suburban trains that feed central intermodal stations in downtown Montreal.

Launched in 2008–2009, the OPUS smartcard fare system is used by transit agencies

throughout the Greater Montreal region (STM, 2015a). OPUS card users purchase a card ($6.00 CAD)

and then load fares onto the card at metro and train stations, as well as authorized vendors throughout

the region. In our study, we focused on transactions from vendors and stations located only on the

Island of Montreal and its associated islands (including Île-Bizard and Île-des-Soeurs) (Fig. 2.1), the

STM’s major service area.

8

Figure 2.1. Context map of the Island of Montreal.

In 2014, a single fare cost $3.00, while a two-way trip fare cost $5.50. Weekly fares (valid from

Monday 5 a.m. to Sunday 11:59 p.m.) cost $24.50, and monthly fares (valid from the first day to the

last day of the month) cost $79.50, so buying weekly passes (four) for an entire month is costlier overall

than purchasing a monthly pass. Moreover, when buying three weekly passes in a month, spending

$6.00 more, or the equivalent cost of two single trips, would have enabled unlimited monthly access;

buying a monthly pass instead of three weekly passes is more worthwhile. While reduced cost senior

and student fare cards are personalized with photo IDs, regular OPUS cards are not personalized.

While we initially wished to study a larger time frame of purchases (six months), the time

requirements to extract transaction data are significant, and thus we limited our data request to a single

month. To our knowledge, no special events occurred in September 2014 that would have significantly

9

affected fare sales, and we chose September to limit the number of potential tourist sales during

summer months or weekly sales because of residents’ holidays during the summer.

2.3.2 Data preparation

A total of 1,010,720 fare purchases were completed within the month of September 2014. These

transactions originated from 602,609 unique OPUS cards and took place both at transit stations (70

locations/Metro or transit stations) and points of sales such as corner stores and pharmacies (363

vendors). Note that these transactions include all cards that made purchases, ranging from single fares,

to monthly fares, to suburban train fares, as well as multiple purchases. Moreover, these transactions

included both regular and reduced fare purchases, which include purchases made by students under

18 years old, students between 18–25 years old, and seniors over 65 years old. For this study, only

regular fares are examined because reduced fares cannot be stratified into different groups (students

vs. seniors) due to the similarity in price. For the five calendar weeks in September 2014, 63,262 regular

weekly passes were sold at transit stations and 34,707 were sold at local vendors. In terms of monthly

passes (both for September and October pass, since monthly passes are available on the 20th day of

the preceding month), 145,970 were sold at transit stations while 57,762 were sold at local vendors.

For data preparation (Fig. 2.2), fare purchases originating from metro stations off the Island of

Montreal were removed because off-island residents need to buy fares for their regions (and were not

recorded in our dataset) as well as fares to access the STM system (present in our dataset). Also

records from the main passenger airport were removed because of the non-residential nature of the

surrounding area and potential skewing due to tourist purchases. Any records that contained more than

one purchase of the same fare type (e.g., two of the same week or two of the same month) were also

removed as errors. In all, transactions taking place at a total of 407 different points of sale

(interchangeably referred to as ‘locations’ including 64 transit stations and 343 local vendors) are

analyzed in this study. The final analyses included 61,959 weekly fares from transit stations (2%

removed) and 32,892 weekly fares from vendors (5% removed). For monthly fare transactions, 144,688

fares were analyzed from transit stations (<1% removed) and 53,394 monthly fares from vendors (7%

10

removed). To analyze recurring purchases of weekly fares (three or more weekly fare purchases during

the month), individual OPUS IDs were extracted and only IDs that made repeated purchases at a single

point of sale (i.e., excluding IDs that bought multiple weekly passes at different locations) were

examined. We also excluded IDs that purchased repeated weekly fares as well as monthly fares. This

subset contained 5,395 unique purchasers of three or more weekly fares at a unique location in the

month of September 2014.

Figure 2.2. Data preparation.

The dependent variables (see Table 2.1) are the total number of monthly fares purchased at a

location, the total number of weekly fares purchased at a location, and the number of purchasers of

three or more weekly fares from a unique location. When testing for normality with Shapiro-Wilks,

Shapiro-Francia and skewness tests, as well as visually analyzing a frequency plot of dependent

variables, we determined that the dependent variables were not normally distributed. Therefore, the

monthly fare dependent variable was transformed as a natural logarithm. Moreover, since five vendors

11

had no weekly sales and 52 vendors had no three or more weekly sales, we used a multilevel Poisson

regression model to account for vendors with dependent variables equal to 0, as well as the non-

normality of the dependent variables.

2.3.3 Income and spatial variables

To link neighbourhood characteristics to each vendor, we generated a 500-metre network buffer

around the centroid of every vendor postal code using the Montreal street network. These buffers were

intersected with census tracts to assign median household income within the buffer around each

vendor, and the contribution of each census tract was weighted based on area. The same procedure

was used for metro stations, but a 1-km network buffer was used instead, assuming the greater draw

of metro stations (El-Geneidy, Grimsrud, Wasfi, Tétreault, & Surprenant-Legault, 2014). Moreover, the

percent of part-time workers was also calculated with the same method since part-time workers may

be prone to purchase weekly passes due to the uncertainty of work scheduling. A similar procedure

was used to calculate population within the buffer. Finally, categorical variables related to the built

environment (metro stations, hub stations, bus stops, commuter stations, and other vendors) were

determined as falling within the network buffer or not, and distance to the downtown center point (CBD)

was calculated using network analyst in a GIS.

To generate the maps in Figure 2.3, individual points of sale were assigned to the census tract

that contributed the greatest area proportion. Then, median income, total monthly fares, total weekly

fares, and total three or more purchasers were normalized by calculating z-scores for each point of

sale. Finally, the z-scores for each variable (by point of sale) were summed by census tract and

displayed as deciles on a map. Hollow census tracts do not contain points of sale. Note that these

maps were used for display purposes, as the unit of analysis is a point of sale and not census tract (or

neighbourhood).

2.3.4 Statistical modelling

To account for the variability in transit fare sales between points of sale located within the same census

tract, we used multilevel mixed-effects Poisson regression modelling when appropriate (tested with a

12

log-likelihood ratio test). Multilevel modelling acknowledges the hierarchical nature of data, such as

individuals nested within different levels like neighbourhoods. Specifically, by accounting for hierarchical

relationships, multilevel modelling meets the assumption of the independence of observations that

ordinary least square regressions may violate. Multilevel modelling permits allocating the variance to

lower (point of sale) and higher (neighbourhood) levels. In other words, if the variance in the data is due

to variance between the neighbourhoods, the multilevel technique reports it in the model, while the

variance from within the neighbourhood is controlled for and does not affect the other reported

variances or the coefficients. The use of this technique is becoming more widespread in social sciences

and has been used recently in several publications in the transport field (Djurhuus, Hansen, Aadahl, &

Glümer, 2014; Familar, Greaves, & Ellison, 2011; Wasfi, Ross, & El-Geneidy, 2013). In addition,

specifying a Poisson model is appropriate for count responses as well as for data with record of no fare

transactions. Note that specifying the models as either linear regressions (ln-transformed), multilevel

linear regressions, or a multilevel mixed-effects Poisson regressions, by and large yielded similar results

in terms of magnitude and sign of the variable of interest, income, demonstrating the stability of the

models.

Each point of sale was assigned to a census tract based on the census tract that made the

largest area contribution from the GIS-generated network buffer. In Montreal, census tracts are valid

proxies for neighbourhoods (Ross, Tremblay, & Graham, 2004). In this study, we used census tracts as

a level to visualize and understand the effect of the individual location and neighbourhood

characteristics on transit fare sales.

2.4 RESULTS

Table 2.1 lists and describes the variables used in this paper, as well as the summary statistics of the

variables. Every point of sale sold an average of 233 weekly passes and 486 monthly passes,

suggesting that over the course of a month, more riders buy monthly passes. In terms of recurring

weekly pass sales, an average of 13 riders purchased three or more weekly passes at every vendor in

13

September 2014. Table 2.1 demonstrates that for both types of fares, standard deviations are high

indicating a large variation in sales among points of sale.

Table 2.1. Description of Variables and Summary Statistics.

Variable name Description Mean Std. Dev. Income Median household income (Canadian dollars) within the buffer

(divided by $10,000) 4.50 1.44

Part-time Percent of workers who are part-time within the buffer 19.77 2.91 Population Population within the buffer (divided by 1,000) 4.02 1.25 Distance to CBD Network distance (km) from vendor or metro station to downtown

center point 9.62 6.00

Distance to CBD^2 Square-term of network distance to downtown 128.39 168.62 Metro (dummy) Dummy variable equal to 1 if the point of sale is a metro station 0.15 0.36 Hub (dummy) Dummy variable equal to 1 if the metro station is a major transfer

station or the main downtown station 0.037 0.19

Metro stations Number of metro stations located within the buffer (excluding the metro station that is the vendor itself)

0.34 0.82

Bus stops Number of bus stops located within the buffer 25.59 26.87 Commuter stations Number of commuter stations located within the buffer 0.057 0.26 Other vendor (dummy) Dummy variable equal to 1 if a vendor (non-metro) is located within

the buffer 0.56 0.50

Dependent variable Total monthly fares Total monthly fares (September and October) sold in the month of

September 486.69 897.27

ln of total monthly fares Natural logarithm of total monthly fares (September and October) sold in the month of September

4.91 1.61

Total weekly fares Total weekly fares (calendar weeks 36, 37, 38, 39 and 40) sold in the month of September

233.05 386.16

IDs 3 or more Number of unique OPUS IDs purchasing 3 or more weekly fares at unique points of sale

13.26 21.21

N = 407 points of sale (64 Metro/transit stations and 343 vendors); 258 neighbourhoods/census tracts

Figure 2.3a shows that the neighbourhoods with the lowest median incomes (blue hues) are

located in the north, southwest, and some central parts of the island. By mapping standardized values

(z-scores) for monthly fare, weekly fare, and recurring weekly fare purchases, it appears that census

tracts with low-incomes had the greatest purchases of all types of fares (red hues, Fig. 2.3b–d). These

maps show that vulnerable neighbourhoods are where the transit fare sales are above the mean,

consistent with the notion that socially vulnerable groups use public transit the most (Giuliano, 2005;

Taylor & Morris, 2015).

14

Figure 2.3. Income and fare purchase deciles. 2.4.1 Transit fare purchase models

Using multilevel linear regression modelling at the census tract-level to account for variation between

vendors and transit stations located within the same census tract, we developed three different models

to determine how different variables contribute to the purchasing of the different fare types. Specifically,

we asked how fare purchases at points of sales depend on median income of the neighbourhood

surrounding it while controlling for a number of spatial characteristics, such as transit stops and

distance to CBD.

15

First, we analyzed total monthly fare sales using a multilevel linear regression, but found that it

was statistically unnecessary (log-likelihood test, P ≥ χ2 = 0.38) and therefore used a regular linear

regression but with a natural logarithmically-transformed dependent variable; this was necessary due to

the non-normality of the dependent variable. This model is based on 198,082 total monthly fare

transactions from 407 points of sale. Table 3.2 shows the output of the model with median household

income as the variable of interest.

Income is not a significant predictor variable of monthly fare purchases, suggesting that

individuals with all levels of income buy monthly fares. Moreover, the percentage of part-time workers in

a neighbourhood does not impact monthly fare purchases. The control variables show that for every

1,000 residents in a neighbourhood, 16% more monthly fares will be purchased, and distance from the

CBD displays a non-linear relationship with monthly fare purchases, so after a certain distance from the

CBD (distance to CBD^2), 0.25% fewer monthly fares are purchased. If the sales location is a metro

station, 239% more monthly fares are sold compared to local vendors, and 1.7% more monthly fares

are purchased for every bus stop within the vicinity of the sales point. Model 1 explains 55% of the

variation in monthly fare purchases.

Table 2.2. Model 1 – Monthly fare purchase model.

Variable Coefficient t-statistic 95% CI Lower

bound 95% CI Upper

bound Income 0.0071 0.15 -0.084 0.10 Part-time 0.00059 0.03 -0.039 0.039 Population 0.16** 2.84 0.049 0.27 Distance to CBD 0.073 1.92 -0.0017 0.15 Distance to CBD^2 -0.0025* -1.98 -0.0051 -0.00002 Metro (dummy) 2.39** 6.79 1.70 3.08 Hub (dummy) 0.57 1.74 -0.075 1.21 Metro stations -0.081 -0.65 -0.33 0.16 Bus stops 0.017** 2.71 0.0046 0.029 Commuter stations -0.26 -1.04 -0.75 0.23 Other vendor (dummy) -0.072 -0.60 -0.31 0.16 Constant 3.11** 5.61 2.02 4.20

N 407 points of sale Adjusted R2 0.55 F statistics (11, 395) 45.25

Prob > F 0.00 AIC 1235.66 BIC 1283.77

Bold indicates statistical significance; ** Significant at 99%; * Significant at 95%

16

Next, we analyzed total weekly fare purchases based on 94,851 transactions from 407 points

of sale. First, we specified a multilevel model using total weekly sales transformed as natural logarithms,

but needed to add 1 to account for points of sale with 0 weekly fare transactions; running the model as

a multilevel was statistically appropriate (log-likelihood test, P ≥ χ2 = 0.0026). To better account for

locations with 0 weekly fare transactions and the non-normality of the dependent variable, we specified

a multilevel mixed-effects Poisson regression. The output of Model 2 is reported in Table 2.3.

Table 2.3. Model 2 – Weekly fare purchase model.

Variable Coefficient z-statistic 95% CI Lower

bound 95% CI Upper

bound Income -0.13** -7.79 -0.17 -0.10 Part-time 0.095** 14.14 0.082 0.11 Population 0.10** 5.58 0.067 0.14 Distance to CBD 0.47** 20.47 0.43 0.52 Distance to CBD^2 -0.015** -16.78 -0.016 -0.013 Metro (dummy) 2.03** 38.21 1.93 2.14 Hub (dummy) 0.63** 34.66 0.59 0.66 Metro stations 0.25** 15.56 0.22 0.28 Bus stops 0.013** 14.49 0.011 0.015 Commuter stations -0.033 -1.76 -0.070 0.0038 Other vendor (dummy) 0.039 1.87 -0.0020 0.080 Constant -0.69* -2.63 -1.21 -0.18 Census tract – Variance 1.03 0.84 1.25

Number of observations 407 points of sale Number of groups 258 census tracts/neighbourhoods

Log-likelihood -6121.97 Prob > χ2 0.00

AIC 12269.93 BIC 12322.05

Bold indicates statistical significance; ** Significant at 99%; * Significant at 95% For every $10,000 increase in median income of the neighbourhood is predicted to decrease

weekly fare purchases by 13%, suggesting that in neighbourhoods with low incomes, weekly fare sales

are greater than in neighbourhoods with high incomes. In addition, every 1% increase in part-time

workers residing in the neighbourhood will be expected to increase weekly fare sales by 9.5%, while

every 1,000 persons in a neighbourhood will increase total weekly fare sales by 10%. For every 1-km

farther from the CBD, there is a 47% increase in weekly fare purchases until a certain distance, where

every additional kilometer reduces sales by 1.5%. This suggests that sales locations in inner-city

neighbourhoods will sell more weekly fares compared to farther, more suburban neighbourhoods.

17

Access to mass transit will also increase the purchasing of weekly fares; if the point of sale is a metro

station, 203% more weekly fares will be purchased compared to purchases at local vendors, and a hub

metro station will sell 63% more weekly fares than local metro stations. Every metro station within the

vicinity of a sales location will increase weekly fare purchases by 25%, while every nearby bus stop

within the vicinity of a sales location will increase weekly fare purchases by 1.3%. As a general linear

model (ln-transformed), Model 2 explains 54% of the variation in total weekly fare purchases.

Finally, we analyzed how recurring weekly fare purchases depend on income. Model 3 (Table

2.4) is based on 5,395 unique purchasers (smartcard IDs) that bought three or more weekly fares

during the month. Similar to Model 2 above, Model 3 was specified as a multilevel Poisson regression

to account for the non-normality of the dependent variables and locations with no purchasers of three

or more weekly fares.

Model 3 reveals that for every $10,000 increase in median household income in a

neighbourhood, 24% fewer recurring weekly fare purchases will be made, suggesting that in

neighbourhoods with low incomes, more transit riders will buy repetitive weekly fares compared to

neighbourhoods with high incomes. Interestingly, while the percentage of part-time workers can

increase total weekly fares (Model 2, above), the same variable is not significant in the recurrent weekly

model, suggesting that part-time workers may purchase weekly fares, but not repetitively to substitute

for a monthly fare. Moreover, for every 1,000 persons in a neighbourhood, 11% more recurrent weekly

purchases will take place. Every kilometre further from the CBD increases recurrent weekly purchases

by 34%, until a given distance, where 0.98% fewer types of these purchases will be made. Recurring

purchases will be made significantly more at metro stations, since metro station sale locations will see

199% recurring weekly fare sales compared to local vendors, and 46% more will occur at hub metro

stations. Interestingly, the presence of a metro station close by will compete for recurring weekly sales

at a vendor of interest, reducing these sales by 16%. The presence of bus stops will also modestly

increase the percentage of recurring weekly pass purchases by 1.8%. As a general linear model (ln-

transformed), Model 3 explains 56% of the variation in purchasers of three or more weekly fares.

18

Table 2.4. Model 3 – Three or more recurrent weekly fare purchase model.

Variable Coefficient z-statistic 95% CI Lower

bound 95% CI Upper

bound Income -0.24** -5.60 -0.32 -0.16 Part-time 0.033 1.84 -0.0022 0.069 Population 0.11* 2.46 0.022 0.20 Distance to CBD 0.34** 8.67 0.26 0.42 Distance to CBD^2 -0.0098** -7.04 -0.012 -0.0070 Metro (dummy) 1.99** 10.65 1.62 2.36 Hub (dummy) 0.46** 5.94 0.31 0.62 Metro stations -0.16* -2.52 -0.29 -0.036 Bus stops 0.018** 5.59 0.012 0.024 Commuter stations -0.025 -0.25 -0.22 0.17 Other vendor (dummy) 0.10 1.44 -0.036 0.24 Constant -1.01 -1.91 -2.05 0.025 Census tract – Variance 0.56 0.44 0.72

Number of observations 407 points of sale Number of groups 258 census tracts/neighbourhoods

Log-likelihood -1379.75 Prob > χ2 0.00

AIC 2785.50 BIC 2837.62

Bold indicates statistical significance; ** Significant at 99%; * Significant at 95% 2.5 DISCUSSION AND CONCLUSION

By modelling transit fare purchase records during September 2014, we discovered that in

neighbourhoods with lower median incomes, fare vendors will sell significantly more weekly fares

overall, as well as have more purchasers of recurring of weekly fares than vendors in neighbourhoods

with higher median incomes. Given that weekly passes in Montreal are more expensive per trip

compared to monthly passes, our findings suggest that low income can spur the sale of repetitive

weekly fares. Moreover, though the proportion of part-time workers influences overall weekly fare

purchases, it is not a significant explanatory variable for recurrent weekly passes, implying that part-time

workers may purchase weekly fares due to scheduling, but they are not driving the sale of recurrent

weekly fares, supporting the idea that low-income individuals substitute monthly fares for multiple

weekly fares (Hickey et al., 2010; Schuerman, 2015). As a consequence, financially vulnerable riders are

more likely to be spending more income over the course of a month than riders who can afford to pay

for the monthly pass in a single payment. Future work should address how transit agencies strive to

balance fares and profits with maximizing ridership and network efficiency.

19

For vulnerable populations, financial affordability is usually the first hurdle to accessing public

transit (Stolper & Rankin, 2016), and access to viable transit for marginalized populations, including

residents who lack private cars (Delbosc & Currie, 2012), is necessary to reach jobs and other

opportunities (Delbosc & Currie, 2011). As transit agencies regularly increase fares (Hickey, 2005), the

burden falls hardest on marginalized groups, in particular when buying monthly fares. The steep upfront

cost of a monthly pass may be unaffordable for low-skilled workers who are paid weekly or biweekly

(Schuerman, 2015); they may opt instead to buy weekly passes for the entire month. If this is indeed

the case, then transit agencies, rather than providing equitable service, are placing an extra penalty on

low-income residents by charging more per ride for weekly fares compared to monthly fares.

Our research shows that monthly sales were not influenced by income. This finding aligns with

previous work in Athens, Greece that found that income had no effect on bus ridership (Gkritza et al.,

2011). In addition, that neighbourhoods in central Montreal show more transit fares sold than suburbs is

expected given that metro and bus availability are confined in central Montreal, while suburbs are

served mainly by commuter rail that require a different fare that was not included or analyzed in this

study. Moreover, sales were largely dependent on whether a point of sale was a metro station or a main

hub, such as a transfer station or the main downtown station; transit agencies should ensure sufficient

numbers of fare vending machines are available at these types of locations.

Transit agencies increase fares to balance budgets and improve service, but this practice may

place extra strain on the neediest riders (Aggarwala, 2012). Since 2013, the STM has increased

monthly fares by $2.50 a year, while weekly fares were increased by $0.75 from 2013 to 2014, and

again by $1.00 from 2014 to 2015; in comparison to monthly fares, this increase was disproportionately

higher for weekly fares. To address fare equity concerns, one potential solution could be to charge the

same amount per ride, so that four weekly passes cost the same total price as a 28-day fare, as in

Washington, D.C. (WMATA, 2015). In London, fares are capped after a maximum specified number of

trips (TfL, n.d.), and having caps that vary by group, like students, seniors or low-income riders, could

potentially address equity concerns related to steep upfront costs. Transit agencies in San Francisco,

20

Seattle, and Calgary are examples of agencies that offer fare subsidizes to low-income riders (SFMTA,

n.d.; King County Transit, 2015; Calgary Transit, 2015) to decrease the financial burden on socially

disadvantaged populations. Using technology already in place with smartcards, the STM could enroll

qualifying low-income individuals to be directly eligible for reduced fares, or alternatively, test fare

capping.

Our research findings highlight the necessity of conducting fieldwork to uncover the purchasing

habits of vulnerable riders. In addition to financial precarity, other vulnerable populations are also transit-

dependent, such as single-parents, recent immigrants, and individuals with low educational attainment

(Foth et al., 2013; Manaugh & El-Geneidy, 2012). Initially, similar to previous equity work related to

accessibility that used an index meant to capture vulnerability at the neighbourhood level (Foth et al.,

2013), we also tested a combined social deprivation index and found that results were similar to the

models using solely income (data not shown). Therefore, our results indicate that other transit-

dependent groups may also purchase multiple weekly fares as a substitute for a monthly fare. Surveys

at ticketing locations in areas with socially marginalized populations would lend credence to our

assumptions and conclusions. Moreover, future work could attempt to model single fare transactions

and its potential relation with social vulnerability. Regardless, a reappraisal of transit fare structure is

warranted, since raising fares for high-income riders who can afford to pay more while subsidizing low-

income riders may help address not only equity concerns (Stolper & Rankin, 2016), but also address

funding shortages and improve transit quality for all (Aggarwala, 2012).

We acknowledge the limitation of analyzing a single month’s worth of purchase data. While we

wished to study a larger time frame, the intensive time requirements for data extraction limited us to one

month’s worth of data. An important limitation of our study is that we used neighbourhood-level

variables as proxies for demographics of fare purchasers. Although the STM uses smartcards, they,

similar to other agencies, do not collect personal information or link such data to smartcard usage

(Pelletier, Trépanier, & Morency, 2011; Utsunomiya, Attanucci, & Wilson, 2006). A pilot study collecting

21

a cross-section of transit riders’ social profile linked to purchase data could confirm the findings in our

study.

22

CHAPTER 3. HOW AFFORDABLE IS QUALITY PUBLIC TRANSPORTATION FOR MINIMUM

WAGE EARNERS IN NORTH AMERICA?

3.1 INTRODUCTION

Urban public transit plays a vital role in society. In addition to providing an important service to residents

on a daily basis, which allows them to access employment and other essential services, particularly for

people who may not own personal vehicles, public transit contributes to reduced motor vehicular

activity, air pollution, energy consumption, and improved road safety, all of which benefit urban

populations at large. For riders who depend on transit, affordability of transit fares is crucial; indeed, a

recent report on the working poor in New York found that fare affordability was the “biggest problem” of

the subway system, ahead of delays and crowding (Stolper & Rankin, 2016). Surprisingly, to our

knowledge, little previous research has focused on the financial affordability of public transit, which is an

important aspect of accessibility, but rather examines equity related to fare subsidies (Hodge, 1988;

Serebrisky, Gómez-Lobo, Estupiñán, & Muñoz-Raskin, 2009). Moreover, public transit agencies in

North America typically face funding shortages (US DOT, 2013) and as such, rely a great deal on fare

revenue for financial viability (TCRP, 1998). Therefore, a tension exists between the ability of public

transit agencies to provide the level of service riders expect while maintaining fares at a socially

acceptable level, which in turn allows transit agencies to remain solvent and provide a viable service

(Badami & Haider, 2007).

We can think of important objectives and outcomes related to urban public transit from the

perspective of, first of all, the travelling public (i.e., transit riders), society at large (since transit provides

public goods of benefit to all of society), and the agencies themselves. In this paper, we ask not only

how affordable is public transit in the largest 14 cities in North America (population more than 3 million

inhabitants), but also how affordability is linked to quality transit service from multiple perspectives. To

this end, we investigated how well transit agencies meet the various objectives important to riders,

23

society and to the agency themselves, and then compare performance to the affordability of this service

to minimum-wage earners. In doing so, this paper uses publicly and freely available data reported by

public transit agencies. The output from this process has the advantage of allowing researchers,

transport planners, as well as interested members of civil society the ability to assess the affordability

and quality of an agency of interest compared to peer agencies without the need for costly and

proprietary surveys (Randall, Condry, & Trompet, 2007).

This paper begins with a brief literature review of previous studies concentrating on transit

quality as it relates to diverse perspectives, as well as transit affordability. The second section of the

paper describes the rationale for each indicator and how the indicators are constructed. In the third

section of the paper, we describe the major findings based on different quality of service indicators in

relation to affordability. In the final section, we discuss the implications of our findings, and how the

methodology for such an assessment may be improved in future work.

3.2 LITERATURE REVIEW

The dilemma of achieving contrasting goals—providing quality service that is affordable, while being

profitable or cost efficient (Badami & Haider, 2007)—requires the acknowledgement of different

stakeholders involved in public transit and their sometimes divergent interests. The Transit Cooperation

Research Program (TCRP) Transit Capacity and Quality of Service Manual provides a primer on quality

of service measures and concepts (TCRP, 2013), and as such, describes four different, and somewhat,

overlapping groups with interests in public transit quality or performance: customer, community,

agency, and motorists who interact with but are not taking public transit (TCRP, 2013). Customer

preferences relate to two broad factors—service availability, and comfort and convenience. Service

availability depends on the frequency of service, on the hours of available service or service span, and

on the accessibility to transit stops or stations (TCRP, 2013). And while transit riders may be choice

riders, implying that these individuals choose to use transit instead of driving a private vehicle, or transit-

dependent riders who have no other option but to use transit (Krizek & El-Geneidy, 2007), comfort and

24

convenience are important to both types of riders and can impact satisfaction and usage (Eboli &

Mazzulla, 2011; Verbich & El-Geneidy, 2016). Comfort can result from the passenger load and seat

availability, the duration of the trip itself, driver friendliness, and reliability of the service. Indeed, a large

body of evidence based on customer satisfaction surveys reiterates the importance of reliability,

frequency, comfort, and safety to transit riders (Das & Pandit, 2016; de Oña, de Oña, Eboli, & Mazzulla,

2013; dell’Olio, Ibeas, & Cecin, 2011; TCRP, 2013). When agencies are able to provide transit service

that closely aligns with rider expectations, transit agencies provide benefits to society at large by

allowing individuals, in particular those without vehicles, to move around a city, as well as help reduce

pollution and congestion.

Indeed, from a societal point of view, an important outcome is the extent to which transit

agencies are able to carry people (APTA, 2016); more particularly, the extent to which they are able to

replace car trips, because that determines the extent to which urban public transit contributes to the

production of public goods (in terms of air pollution, congestion, road safety, etc.). From the view point

of an agency, and particularly if the agency is funded through taxpayer revenues, it is important that

transit contributes to the achievement of the objectives and outcomes of importance to commuters and

society at large, while also being financially viable (TCRP, 2013). Indeed, urban transit agencies face

what might be considered a coverage-affordability-viability dilemma—that is, there is a potential trade-

off between their ability to provide widespread coverage, and to do so affordably for commuters, while

also being financially viable.

A well-used transit service is beneficial to society at large, certainly the agency, and riders as

well; therefore, many features valued by one group may also be important to another, and are difficult to

unequivocally classify as one group's interest over another's. While passengers may be unaware or

unconcerned with aspects like cost effectiveness or ridership of a route, for instance, these measures

are priorities for agencies and directly impact the service quality riders’ experience. Operating revenue

for transit service is derived from a variety of sources, such as government funding, advertisement and

other revenues, and importantly, fare revenues. Fare revenue makes up 25% of the total budget of

25

agencies in regions with populations greater than 200,000, and fares contribute 33%—the largest

single source—to operating expenses (TCRP, 2009). Not surprisingly then, transit agencies rely a great

deal on fares, and these fares have outpaced inflation in the United States between 1989–1994 (TCRP,

1998). In Canada, limited federal funding means that for transit agencies, operating expenses are

mostly covered through provincial funding and locally generated revenues and fares. For example, in

Montreal, 41% of the 2016 operating budget (a little over $1.5 billion (CAD)) of the Société de transport

de Montréal (STM) comes from fares, while 34% comes from local and regional governments, while

nearly 23% is provided by the provincial government, and about 2% coming from other sources (STM,

2015b). In the United States and Canada, farebox revenue is thus a crucial budget item. This significant

dependence on transit fares, coupled with increasing expenditures, forces agencies to attempt to

maximize revenue but maintain fares at an acceptable level—a viability-affordability dilemma (Badami &

Haider, 2007). While raising transit fares may be necessary for transit viability, fare increases may also

overburden low-income groups as many of them rely on public transit as a main mode of transport

(APTA, 2007; Stolper & Rankin, 2016).

Affordability studies of public transit have been scarce. In a Word Bank publication, Carruthers,

Dick, and Saurkar (2005) derived a simple and useful metric to assess transit affordability among

developing and developed countries. This affordability index assumes 60 ten-kilometre trips (single

fares) per month expressed as a percentage of average annual per capita income of a city. On this

basis, they determined that residents of Latin American cities spend, on average, between 4–11% of

annual income on transit, while residents of Western cities spend, on average, less than 5%. They also

repeated this exercise for the bottom quintile earners and discovered that residents in Western cities

spend upwards of 10% of income on transit, while residents in Latin American cities spend more that

25% of their annual income on transit (Carruthers et al., 2005). Recent work has used Gini coefficients

based on the cumulative distribution of transport benefits to determine how different groups, such as

students or the elderly, benefit or lose based on different subsidy policies for transit fares (Serebrisky et

al., 2009). Nonetheless, deciding upon an acceptable level of income devoted to transport is not trivial

26

(Gómez-Lobo, 2011). Therefore, beyond the importance of the quality of public transit performance for

riders, society, and agencies, affordability of public transit is also crucial, particularly for vulnerable

residents (Stolper & Rankin, 2016).

Based on this overview of the literature, it is clear that multiple perspectives are involved in

public transit outcomes and an intricate balance is needed to achieve sometimes conflicting goals,

namely quality service for an affordable price. Figure 3.1 schematizes the relationship between the fares

that transit riders pay which play a role in determining the service quality provided by a transit agency,

and how the service quality can determine the transit fare charged by an agency. As such, a transit

agency needs to balance the quality of factors like accessibility, frequent service, and comfort with

affordability. In the end, society accumulates multiple benefits from a viable public transport service.

Figure 3.1. Balancing affordability and transit service quality from multiple perspectives. 3.3 METHODOLOGY

The current paper is concerned with the public transport service residents in large (3 million inhabitants

and more) North American cities have access to. Importantly, this paper is not a comprehensive

27

evaluation or benchmarking study of transit agencies per se, but rather aims to assess how transit

service quality may vary with transit affordability.

Determining a peer group for transit agency comparative studies is not a trivial task, and the

research goal should inform inclusion and exclusion criteria. First, we limited the agencies included in

this analysis to the main transit service providers of the largest North American cities, those with

metropolitan populations greater than 3 million inhabitants. It is important to note that the population of

the core city itself may be actually quite small. We also restricted our analysis to agencies that operate

at least two modes, namely bus and rail, including light rail, heavy rail (metro or subway), and/or street

rail (cable car, streetcar, etc.). Since the main concern of this study is urban transit, we excluded

agencies providing commuter rail service only and commuter rail data from agencies included in this

study. In total, we analyzed data from 14 transit agencies, including two from Canada and 12 from the

United States. Table 3.1 lists, in order of decreasing metropolitan population, the cities and agencies

examined in this paper.

Table 3.1. Cities and transit agencies. Metropolitan area Core city Metropolitan

population Transit Agency Modes

New York-Northern New Jersey-Long Island, NY-NJ-PA

New York 18,351,295 MTA New York City Transit (NYCT)

Heavy rail, bus

Los Angeles-Long Beach-Santa Ana, CA

Los Angeles 12,150,996 Los Angeles County Metropolitan Transportation Authority (LACMTA)

Heavy rail, light rail, bus

Chicago-Joliet-Naperville, IL-IN-WI

Chicago 8,608,208 Chicago Transit Authority (CTA)

Heavy rail, bus

Toronto (Mississauga), ON Toronto 5,521,235 Toronto Transit Commission (TTC)

Heavy rail, light rail, street rail, bus

Miami-Ft. Lauderdale-Pompano Beach, FL

Miami 5,502,379 Miami-Dade Transit (MDT)

Heavy rail, bus

Philadelphia-Camden-Wilmington, PA-NJ-DE-MD

Philadelphia 5,441,567 Southeastern Pennsylvania Transportation Authority (SEPTA)

Heavy rail, street rail, bus

Dallas-Fort Worth-Arlington, TX

Dallas 5,121,892 Dallas Area Rapid Transit (DART)

Light rail, bus

Houston-Sugar Land-Baytown, TX

Houston 4,944,332 Metropolitan Transit Authority of Harris County (Metro)

Light rail, bus

Washington-Arlington-Alexandria, DC-VA-MD-WV

Washington 4,586,770 Washington Metropolitan Area Transit Authority (WMATA)

Heavy rail, bus

Atlanta-Sandy Springs-Marietta, GA

Atlanta 4,515,419 Metropolitan Atlanta Rapid Transit Authority (MARTA)

Heavy rail, bus

28

Metropolitan area Core city Metropolitan population Transit Agency Modes

Boston-Cambridge-Quincy, MA-NH-RI

Boston 4,181,019 Massachusetts Bay Transportation Authority (MBTA)

Heavy rail, light rail, bus

Montreal (Laval), QC Montreal 3,752,475 Société de transport de Montreal (STM)

Heavy rail, bus

San Francisco-Oakland-Fremont, CA

San Francisco

3,281,212 San Francisco Municipal Railway (SFMTA)

Light rail, street rail, bus

Seattle-Tacoma-Bellevue, WA

Seattle 3,059,393 King County Department of Transportation (King County Metro)

Street rail, bus

After selecting peer agencies, we next turned to choosing key indicators meant to capture the

experience and service that matter to riders, operators, and society. As far as riders are concerned,

outcomes that are crucially important are availability and accessibility; frequency and reliability; comfort

and convenience; and affordability (p. 9, RPA (2015)). To put it succinctly, riders are looking for high

service quality, in terms of the first three sets of outcomes, but which is also affordable. Unfortunately,

comparable data to capture accessibility, and frequency and reliability, are extremely difficult to obtain

or to generate for all of the metropolitan areas that we have chosen for this paper; these measures will

need to be developed in the future. For instance, frequency and reliability (related to waiting time) are

two chief concerns of transit riders (Diab, Badami, & El-Geneidy, 2015) and tracked by many agencies

(New York City Transit and Bus Committee, 2016; STM, 2016; TTC, 2016), but agencies do not adhere

to a uniform measure of reliability making comparisons difficult, in addition to not reporting

disaggregated statistics. At present, we used publicly available data to come as close as possible to

approximating four main features that are measurable and important to riders—accessibility, service

frequency, comfort, and affordability.

Travel on transit, or any other mode for that matter, is typically a derived demand for access to

some desired destination. As such, accessibility measures provide a good indication of the potential

ability to reach desired destinations, like jobs, health services, or schools, based on a travel mode and

influenced by land use (El-Geneidy & Levinson, 2006; Geurs & van Wee, 2004). Therefore, we used

accessibility measures to jobs by transit within a 30-minute travel time threshold to assess access to

jobs. Values from American cities were from the Access Across America report (Owen & Levinson,

2014), while accessibility in Canadian cities were derived from previous work (El-Geneidy et al., 2015;

29

El-Geneidy et al., 2016). Note that accessibility data from Owen and Levinson (2014) were calculated

using all public transit agencies in a region, while our focus here is the main transit agency of a city; the

agencies analyzed here, as the main agency in each city, likely contribute the greatest to the

accessibility measures used.

To approximate service frequency, we downloaded General Transit Feed Specification (GFTS)

data for all agencies for a date in fall 2014 in order to avoid summer or winter holiday scheduling and

calculated scheduled headways. Headways for all modes operated (≥60 s) were calculated at the stop-

level for stops scheduled from 8:00 a.m. to 8:59 a.m. and from 9:00 p.m. to 9:59 p.m. for weekdays.

Next, 500 random headways (or fewer if 500 stops were not made) were extracted for the two time

periods and average headway was calculated for each time period. While these values are

approximations of actual headways, they nevertheless may provide an indication of service frequency at

two different times of the day.

As for comfort and convenience, load factor—the ratio of passenger kilometres to carrying

capacity kilometres, expressed as a percentage, with capacity kilometres being obtained by multiplying

the total seating and standee capacity of all buses by the kilometres operated—would be a good

measure for this purpose. Unfortunately, load factor data (or the total capacity on the transit fleet) is not

available for the various agencies in North America, so we chose to estimate a proxy measure of

crowding on transit vehicles of all modes (and thus, the space available, and comfort and convenience

for commuters), which is the ratio of passenger kilometres to revenue kilometres. Note that, while from

a rider perspective, a greater value of this ratio is undesirable since it indicates less room on transit, a

high figure might be good from the agency perspective.

A previous study from the World Bank Group examined the affordability of public transit in

developing countries (Carruthers et al., 2005), and provided a snapshot of the amount of income

dedicated to transit for the average resident of a city. Here, we are interested in determining the

affordability for transit-dependent riders, in particular those who earn a minimum wage. These residents

would likely rely on public transit (APTA, 2007) and we assumed that a monthly pass would be the most

30

frugal fare to purchase if one commutes daily, as well as to enable comparability between agencies.

Thus, we used monthly fare (in 2014) and minimum wage (in 2014) in local currency and developed an

indicator to express the number of minimum wage hours needed to purchase a one-month unlimited

fare (fare divided by hourly minimum wage). Taxes on the fare and on income were not included in the

calculations.

For society as a whole, as discussed earlier, an important outcome is the extent to which transit

agencies are able to carry people, as a proportion of the total population, but more particularly, the

extent to which they are able to replace car trips; so, an ideal measure for this purpose would be total

transit trips divided by the product of the population in the service area and the car ownership rate.

However, since car ownership rate is not available in a disaggregated fashion for all of the cities in our

analysis, we use total unlinked trips divided by the service area population as our measure. Finally, the

measure we chose to capture the key outcome from the agency’s perspective is the farebox recovery

ratio—that is, the share of operational costs recovered through farebox revenues.

All data except for headways were acquired through the National Transit Database (NTD) for

American agencies and from the Canadian Urban Transit Association (CUTA) for Canadian agencies.

Data were collected disaggregated by mode to avoid collecting statistics for modes other than bus,

heavy rail (metro or subway), light rail, and/or street rail; data by mode were then summed. Minimum

wages for American cities were acquired through a website listing wages by city and state (Dolye,

2015), and verified by checking with other sources, while Canadian minimum wages were acquired

through a Canadian governmental website (Government of Canada, 2016).

To address the different perspectives involved in public transit, we borrow from multicriteria

decision making (MCDM) analysis, since this approach is amenable to evaluating complex issues

involving different stakeholders with conflicting objectives, as well as providing a transparent process.

Importantly, MCDM analysis compels one to clarify goals and outcomes, and as such, develop

indicators or metrics that are measurable, objective, and reliable (Hobbs & Horn, 1997; Ingram, 2009;

Manaugh et al., 2015). For the present study, we take the MCDM approach to analyze outcomes

31

important to transit riders, society at large, and transit agencies. When combining individual metrics,

such as frequency and FRR, we normalized the values for each agency by using the form of a single

attribute utility function and by assuming linearity of this function over its range between the worst and

best values (Hobbs & Horn, 1997).

The accessibility indicator is the normalized average number of jobs reachable by public transit

within a 30-minute travel time threshold (weighted by number of workers in each census tract or block)

and the greater the number of jobs the better. A combined headway indicator was generated by

summing the two individual normalized values of headways at 8 a.m. and 9 p.m. and then dividing by

two and rescaling; a larger value of this indicator indicates better (shorter) headways. The comfort

indicator is simply the normalized value of total passenger kilometres divided by total revenue

kilometres; larger values of this indicator suggests fewer passengers per revenue kilometre or more

space for a rider and is ‘better’. A final composite indicator, called the rider indicator, was composed by

summing the accessibility indicator, the headway indicator, and the comfort indicator, and dividing by

three, and finally rescaling. Note that none of the indicators were weighted when calculating this

composite indicator. The society indicator is the normalized value of the total unlinked ridership divided

by service area population, where larger values are ‘better’. The agency indicator is the normalized

FRR, which was calculated by dividing total fare revenue by total operating expenses for only modes

analyzed here; greater FRR is 'better'. In the graphs below, normalized values are plotted only when

indicators are combined, otherwise, raw values are plotted.

3.4 RESULTS AND DISCUSSION

3.4.1 Rider perspective

Table 3.2 lists raw values of the indicators aimed at capturing the essence of riders’ desires in a public

transit service. Table 3.2 also lists the normalized values of each indicator, calculated with a form of a

single attribute utility function (Hobbs & Horn, 1997), which were subsequently used to derive a

32

composite rider indicator. Agencies are listed from best to worst performing from the composite rider

indicator, which is an unweighted average of the accessibility, headway, and comfort indicators.

Table 3.2. Transit rider indicators.

City Transit agency

Ride

rshi

p (u

nlin

ked

trip

s,

‘000

)

Avg.

ac

cess

ibili

ty

(jobs

)

Acce

ssib

ility

in

dica

tor

(a)

Avg.

8 a

.m.

head

way

(min

)

Avg.

9 p

.m.

head

way

(min

)

Hea

dway

in

dica

tor

(b)

Avg.

pas

s km

/rev

. km

Com

fort

in

dica

tor

(c)

Rider indicator

(a+b+c/3)

New York NYCT 3,631,168 210,186 1.00 8.55 13.86 0.73 28.53 0.00 1.00

Chicago CTA 466,658 48,116 0.20 8.84 13.57 0.73 17.31 0.62 0.81

Toronto TTC 514,216 60,676 0.26 5.80 7.48 1.00 23.48 0.28 0.81

San Francisco

SFMTA 857,806 65,246 0.29 9.06 13.77 0.71 21.94 0.36 0.62

Boston MBTA 99,108 49,237 0.21 8.18 14.39 0.73 20.89 0.42 0.61

Philadelphia SEPTA 301,146 35,317 0.14 8.99 19.57 0.56 16.85 0.65 0.60

Houston Metro 66,842 15,166 0.04 15.24 21.85 0.28 10.47 1.00 0.57

Seattle King County Metro

72,694 26,141 0.09 13.04 17.38 0.48 15.11 0.74 0.57

Montreal STM 409,197 70,683 0.31 8.79 12.97 0.75 23.94 0.25 0.57

Miami MDT 128,539 15,333 0.04 10.87 17.96 0.54 16.55 0.66 0.49

Washington, D.C.

WMATA 359,715 47,759 0.20 13.94 20.20 0.37 17.08 0.63 0.45

Dallas DART 698,473 10,113 0.02 19.89 26.22 0.00 10.97 0.97 0.23

Los Angeles LACMTA 155,317 43,430 0.18 12.72 18.86 0.45 22.26 0.35 0.21

Atlanta MARTA 101,352 6,995 0.00 20.90 20.77 0.12 16.72 0.65 0.00

Average – 50,307 – 11.78 17.06 – 18.72 – –

Standard deviation – 50,480 – 4.48 4.72 – 5.05 – –

Coefficient of variation – 1.00 – 0.38 0.29 – 0.27 – –

Accessibility to jobs by transit helps explain the efficiency of a transit system to deliver workers

to jobs—a critical role of public transit (Foth et al., 2013). Accessibility values during the morning peak

to jobs by transit within a 30-minute threshold were derived from Access Across America, and

calculated for Toronto and Montreal by weighting job accessibility by the number of workers (per

census tract), and averaged over the metropolitan region. We caution that because of different

geographical scales and methods (average accessibility over 7–9 a.m. for American cities vs.

accessibility at 7 a.m. for Canadian cities), and since these measures encompass all transit modes

(including commuter rail), the accessibility measures likely overestimate the accessibility of the particular

agencies examined here. Regardless, they provide a useful comparison (Table 3.2). New York’s transit

33

system offers the largest accessibility to jobs, while Montreal’s transit agency, the STM, offers the

second best accessibility to jobs. Atlanta’s transit agency, MARTA, offers its residents the lowest

accessibility to jobs in the cohort. Unsurprisingly, headways are shorter in the morning peak than in the

evening hours. The TTC has the shortest average headways during both time periods, while MARTA

has the longest headways during the morning peak, and DART has the longest headways at off-peak

evening hours (Table 3.2). Interestingly, most agencies with short headways perform poorly on the

comfort indicator, such that the TTC, for instance, has 23.48 passenger kilometres per revenue

kilometres, while Houston's transit agency, Metro, DART, and MARTA have long headways, but few

passenger kilometres per revenue kilometres. These three individual indicators were normalized,

averaged in an unweighted manner, and finally rescaled to yield the composite rider indicator and

ranked from best to worst performing (Table 3.2). NYCT, the CTA, and TTC are the top performing,

while DART, the LACMTA, and MARTA at the bottom performing transit agencies from a rider

perspective.

Table 3.3 reports the cost of a monthly fare, hourly minimum wage, the number of hours of

minimum wage work needed to purchase a transit fare, as well as a normalized affordability indicator

(fewest hours needed to afford a monthly fare is 'best' or 1.00). Note that values are in local currency

and from 2014. The transit agencies with fares that are most affordable for minimum wage earners are

the SFMTA, the STM, and the LACMTA. The least affordable agencies—those with fares that require

most hours of minimum wage work—are NYCT, MDT, and WMATA.

Table 3.3. Transit agencies, fares, minimum wages, and affordability.

City Transit agency Monthly fare ($) (a)

Hourly minimum wage ($)

(b)

Hours to purchase fare

(a/b)

Affordability indicator

San Francisco SFMTA 68.00 10.74 6.33 1.00

Montreal STM 79.50 10.35 7.68 0.93

Los Angeles LACMTA 75.00 9.00 8.33 0.89

Seattle King County Metro

81.00 9.32 8.69 0.87

Boston MBTA 75.00 8.00 9.38 0.84

Dallas DART 80.00 7.25 11.03 0.75

Chicago CTA 100.00 8.25 12.12 0.69

34

City Transit agency Monthly fare ($) (a)

Hourly minimum wage ($)

(b)

Hours to purchase fare

(a/b)

Affordability indicator

Toronto TTC 133.75 11.00 12.16 0.69

Houston Metro 90.00a 7.25 12.41 0.67

Philadelphia SEPTA 91.00 7.25 12.55 0.67

Atlanta MARTA 95.00 7.25 13.10 0.64

New York NYCT 112.00 8.00 14.00 0.59

Miami MDT 112.50 7.93 14.19 0.58

Washington WMATA 237.00b 9.50 24.95 0.00

Average 11.92

Standard deviation 4.48

Coefficient of variation 0.388 a The Houston transit agency does not have a monthly fare, so the maximum daily unlimited fare ($3.00) was simply multiplied by 30 days. b While WMATA has an array of monthly fares available by mode, maximum price of a trip, etc., this fare was chosen to be comparable with other agencies, i.e., offering unlimited travel for a 30-day period.

In Figure 3.2, we plotted the different indicators from Table 3.2 against the affordability indicator

from Table 3.3 to visualize the interplay between affordability and service quality attributes. Note in the

following figures, marker sizes represent unlinked passenger trips to provide an understanding of the

volume of passengers handled by each system.

In terms of accessibility, New York’s transit system (NYCT, agency 1) offers the highest

accessibility to jobs at a slightly unaffordable cost (Fig. 3.2a). Montreal’s transit agency, the STM

(agency 12), offers the second best accessibility to jobs and is also the second most affordable agency

in the cohort. While WMATA (agency 9) offers relatively good accessibility, it is the least affordable of the

cohort. Overall, the most affordable agencies tend to offer low accessibility.

In Figure 3.2b, many agencies have shorter headways with affordable service, including

agencies in Boston, San Francisco, Montreal. On the other hand, Atlanta’s transit agency is expensive

despite having longer headways, given that 13.10 hours of minimum wage work is required to purchase

a monthly pass, compared to the average 11.92 hours of the sample. Washington, D.C.’s transit

agency, WMATA, has relatively long headways and is the least affordable of the agencies analyzed

here. To make matters worse, due to increasing operating costs outpacing operating revenue, WMATA

35

has recently considered increasing fares, as well as increasing already long headways to maintain fiscal

viability (Duggan, 2014).

Figure 3.2. Accessibility, headway, and comfort indicators plotted against a measure of affordability. Note that only headway is normalized since this indicator is composed of two measures (headways at peak and off-peak hours). In addition to frequent service, being comfortable on-board is also important for riders (dell’Olio

et al., 2011; Diab et al., 2015). As shown in Figure 3.2c, DART (agency 7) and Metro (agency 8) have

few passenger kilometres per revenue kilometres, suggesting that riders will be relatively more

comfortable than on transit in New York (agency 1). DART and Metro also offer somewhat affordable

transit. Overall, systems with high ridership, indicated by the size of the markers, offer lower comfort

36

that may be affordable, like in Montreal and San Francisco, or more expensive like in New York and

Chicago.

3.4.2 Society and transit agency perspectives

From a societal perspective, a well-used public transit system can be ascertained by analyzing ridership

per capita, an indication of the usage of a particular transit system due to quality service (TLS, 2014).

Table 3.4 lists the raw and normalized values for ridership per capita and ranks them from greatest to

least. The top three used systems, on a per capita basis, are NYCT, the STM, and the TTC. The least

used systems are those found in historically car-dependent cities in the United States, such as in Los

Angeles, Miami, Dallas, and Houston. In particular, transit in Los Angeles has low usage despite its

relative affordability for minimum wage earners (third most affordable) and investment and rail expansion

(Linton, 2016; Nelson & Weikel, 2016). This low per capita usage for LACMTA is likely related to long

headways and poor job accessibility, and can explain the high comfort indicator (see Fig. 3.2).

Table 3.4. Ridership per capita and society indicator. City Transit agency Ridership per capita Society indicator New York NYCT 427.65 1.00

Montreal STM 360.41 0.84

Toronto TTC 305.43 0.70

San Francisco SFMTA 192.88 0.42

Chicago CTA 150.09 0.32

Washington WMATA 110.01 0.22

Philadelphia SEPTA 89.76 0.17

Boston MBTA 86.04 0.16

Atlanta MARTA 79.71 0.14

Los Angeles LACMTA 54.09 0.08

Seattle King County Metro 49.57 0.07

Miami MDT 39.70 0.05

Dallas DART 27.42 0.02

Houston Metro 19.67 0.00

Average 142.32 – Standard deviation 131.34 – Coefficient of variation 0.92 –

For the long-term ability to serve rider and society objectives, it also necessary that transit

service be financially viable or else fiscal issues can arise and service may suffer. A common measure of

37

financial efficiency is FRR, and Table 3.5 lists FRR and normalized values (agency indicator), ranked

from agency with best recovery to agency with the worst recovery. Interestingly, the two Canadian

agencies in the sample, the TTC and STM, are the best performing for FRR, as the TTC recovers 71%

of its revenue from fares, while the STM recovers 56%. NYCT is close in third place at 51% recovery.

The lowest performing agencies for FRR are in southern American cities, including the LACMTA, DART

and Metro. Indeed, funding for Metro (in Houston) comes largely through sales tax revenue (METRO,

2015). Moreover, agencies with high FRRs are those with high ridership, short headways, and good

accessibility, but not necessarily the highest fares; while the TTC is in the middle for affordability, the

STM is second most affordable and has the second best FRR, while NYCT and WMATA are expensive

but have the third and fourth best FRRs, respectively.

Table 3.5. Farebox recovery ratio (FRR) and agency indicator. City Transit agency FRR Agency indicator Toronto TTC 0.71 1.00

Montreal STM 0.56 0.75

New York NYCT 0.51 0.67

Washington WMATA 0.48 0.62

Chicago CTA 0.44 0.55

Boston MBTA 0.40 0.49

Philadelphia SEPTA 0.36 0.41

Seattle King County Metro 0.33 0.36

Atlanta MARTA 0.31 0.33

Miami MDT 0.28 0.27

San Francisco SFMTA 0.26 0.25

Los Angeles LACMTA 0.25 0.23

Dallas DART 0.15 0.06

Houston Metro 0.12 0.00

Average 0.37 – Standard deviation 0.16 – Coefficient of variation 0.44 –

3.4.3 Overall service quality

To determine the overall quality from the rider perspective and how it relates to affordability, we

averaged all the individual indicators—accessibility, headway, and comfort—and rescaled the average

value to derive a service quality indicator pertinent to riders (rider indicator) and plotted it against

38

affordability (Fig. 3.3a). This represents the key trade-off from the perspective of transit riders who

desire quality service that is affordable.

The majority of agencies analyzed here offer good quality service that is somewhat affordable

(Fig. 3.3a). For instance, Chicago’s agency, the CTA (agency 3), and the TTC (agency 4) in Toronto

offer excellent service that is somewhat affordable, while the most affordable agency, SFMTA (agency

13) in San Francisco, also provides good service. The second most affordable agency, the STM

(agency 12) in Montreal, provides somewhat good quality service comparable to agencies in

Philadelphia (agency 6), Houston (agency 8), and Seattle (agency 14), but that is more affordable for

minimum wage earners. Also, while transit in Philadelphia, Houston, Chicago, Toronto, and Atlanta have

comparable affordability, service quality varies drastically. On the other hand, transit in Los Angeles

(agency 2), Dallas (agency 7), and Atlanta (agency 10), while being somewhat affordable, have the

lowest service quality for riders in the cohort. Finally, WMATA (agency 9) in Washington, D.C. is

expensive compared to all other agencies and the quality of service is in the mid-lower section in terms

of rider expectations.

From a societal viewpoint, the STM (agency 12, Fig. 3.3b) provides the second most ridership

per capita that is also second most affordable. The TTC and NYCT too provide significant public

benefits but at a higher cost to minimum wage earners. Overall, a sizeable number of agencies cluster

in the lower left-hand portion in Figure 3.3b, suggesting that while agencies provide relatively affordable

service, they offer low ridership per capita (also see Table 3.4).

The top performing agency from a financial standpoint is the TTC in Toronto, and is seventh

most affordable, while the STM in Montreal is the second best agency at recovering revenues from

fares while being second most affordable (Fig. 3.3c). The agency that is most affordable, SFMTA, ranks

low (11th) in FRR. While NYCT is third best at recovering revenues from fares and WMATA in

Washington, D.C. ranks fourth, they are the least affordable agencies (12th and 14th, respectively).

Interestingly, MDT in Miami, is 10th for the agency indicator, and is the 13th most affordable agency,

suggesting that it is both expensive and financially precarious (Fig. 3.3c). Many of the agencies analyzed

39

here are either expensive and recover a low amount of revenue from fares (Metro, WMATA, MARTA, for

example), expensive and have a higher FRR (TTC and NYCT), or rather affordable but with a low FRR

(SFMTA and LACMTA, for example); the STM in Montreal stands out as a financially viable system with

an affordable fare for minimum wage earners.

Figure 3.3. Rider, society, and agency indicators plotted against a measure of affordability.

Finally, we summed the rider indicator, the society indicator, and the agency indicator, averaged

and re-normalized the values to obtain a service quality indicator listed in Table 3.6. The agencies are

ranked from best to worst overall performing, and the affordability indicator (minimum wage hours

needed to afford a monthly fare) is also listed for comparison and convenience. Moreover, Figure 3.4

40

represents this relation graphically. The top performing agency is NYCT, based on excellent rider and

society indicators; however, buying a monthly fare requires 14 hours of minimum wage work. The TTC

ranked second based on strong rider and agency indicators, and requires 12 hours of minimum wage

work to afford a monthly fare. In third place, the Montreal transit agency, the STM, performs well in the

society and agency indicator, while is about average for the rider indicator. Nevertheless, it is the

second most affordable (~8 hours of minimum wage work to purchase a monthly fare) out of the top-

five performers; only the SFMTA is more affordable and placed fifth. The CTA in Chicago, with a strong

rider indicator, ranks fourth but is roughly equivalent to the TTC in terms of affordability. Finally, out of

the bottom three performers, save for MARTA in Atlanta, both the DART in Dallas and the LACMTA in

Los Angeles offer relatively affordable service, but that is of lower quality. Therefore, from the view of

quality and affordability, the STM and SFMTA are the best performers from the cohort.

Table 3.6. Final ranking, service quality indicator and affordability

City Transit agency Rider indicator

Society indicator

Agency indicator

Service quality

indicator

Hours to purchase

fare New York NYCT 1.00 1.00 0.67 1.00 14.00

Toronto TTC 0.81 0.70 1.00 0.86 12.16

Montreal STM 0.57 0.84 0.75 0.74 7.68

Chicago CTA 0.81 0.32 0.55 0.56 12.12

San Francisco SFMTA 0.62 0.42 0.25 0.43 6.33

Boston MBTA 0.61 0.16 0.49 0.37 9.38

Washington WMATA 0.45 0.22 0.62 0.36 24.95

Philadelphia SEPTA 0.60 0.17 0.42 0.35 12.55

Seattle King County Metro 0.57 0.07 0.36 0.27 8.69

Miami MDT 0.49 0.05 0.27 0.20 14.19

Houston Metro 0.57 0.00 0.00 0.14 12.41

Los Angeles LACMTA 0.21 0.08 0.23 0.07 8.33

Atlanta MARTA 0.00 0.14 0.33 0.03 13.10

Dallas DART 0.23 0.02 0.06 0.00 11.03

41

Figure 3.4. Service quality indicator plotted against a measure of affordability.

3.5 CONCLUSIONS

In this assessment, we distilled the qualities desired by transit riders, society, and transit operators into

simple and easily obtainable indicators to assess the quality and affordability of transit agencies in large

North American cities. We discovered that generally, agencies that offer quality service are expensive

for minimum wage earners, while some of the lower overall performers are generally less expensive,

while some poor performers are actually quite expensive. Nevertheless, two exceptions stand out, the

STM in Montreal and the SFMTA in San Francisco, that are able to provide good service that is the

most affordable in the cohort analyzed here.

Our study does have limitations and our indicators caveats. One limitation is the use of data

from a single year, rather than studying potential changes in service quality over time. Indeed, in 2015,

Metro in Houston reviewed its bus network and is now running new bus routes with greater frequencies

(TransitCenter, 2016). Therefore, using the indicators developed here, Houston's ranking could change

depending on whether the service adjustments and routes improve the 'headway' and 'accessibility'

indicators. Therefore, our indicators are highly relevant in this context, and cases like Houston's

highlight the need to study changes in quality (and affordability) over time. Moreover, our indicators are

42

simplifications meant to address the qualities of public transport important to riders, society, and

agencies. Unfortunately, metrics like reliability, travel time, on-board comfort, and the use of information

technologies, go unreported or are non-standard across the industry. Take, for example, ‘reliability’,

which is defined by different agencies as on-time performance, excess wait time, or headway

adherence, among others (TCRP, 2013; Trompet, Liu, & Graham, 2011). Clearly, standardization and

reporting can help in future assessments of transit quality. Moreover, choosing different indicators may

result in a different ranking.

Using hourly minimum wage and the cost of an unlimited monthly fare, we devised a simple

metric aimed at capturing the amount of work hours necessary to afford a monthly fare. Previous work

has used single fares to estimate affordability, because apparently, this fare is the most accessible to

low-income individuals (Carruthers et al., 2005). Without knowing the total salary of individuals earning

minimum wage, it is hard to gauge what is an acceptable number of hours devoted to earning a transit

fare. Nevertheless, transit fares are a chief concern of the working poor (Carruthers et al., 2005),

particularly given the low minimum wages in many American cities. In some cities, such as Seattle and

San Francisco, additional financial subsidies are available to low-income earners and may be a suitable

strategy for improving public transport affordability (Stolper & Rankin, 2016), as well as subsidies offered

through employers.

This paper aimed to provide a framework for assessing transit agencies and their affordability

based on publicly available data. With this framework, researchers and planners can evaluate and

locate areas of weakness or strength, as well as determine whether quality service necessitates

expense. In addition, with a clear methodology that can be easily adopted, any agency around the

world can calculate similar indicators and benchmark against our current findings. Future studies could

use a comparable methodology to evaluate changes over time to learn whether agencies are improving

or worsening, and how these conditions may impact affordability and attractiveness of service.

Furthermore, our primary focus was on qualities important to riders like affordability, accessibility,

service frequency, and comfort; however, not all riders value these qualities to the same extent. For

43

instance, transit-dependent or captive riders are influenced less by comfort (Krizek & El-Geneidy, 2007),

but affordability can be a major concern (Stolper & Rankin, 2016). Therefore, using customer surveys

can help weight and identify indicators relevant to particular rider segments when composing a

composite indicator.

44

CHAPTER 4. AFTERWORD AND CONCLUSIONS

Public transit can only benefit those who are able to access it—physically and financially. While previous

research has investigated the social justice implications of spatial access to public transit and the

destinations they serve (El-Geneidy et al., 2015; Geurs et al., 2009; Manaugh et al., 2015), there is a

dearth of knowledge regarding equity and transit fares and affordability (Serebrisky et al., 2009). In this

project, I studied how income predicts transit fare purchases, as well as the relation between

affordability and the quality of public transit.

Using fare purchase records from metro station vendors and local neighbourhood vendors, I

modelled monthly, weekly, and multiple weekly fare sales to determine the dependence of each on

sociodemographic and built environment characteristics of the surrounding neighbourhood. Essentially,

I discovered that increases in neighbourhood income would be expected to decrease the sales of

multiple weekly fares at vendors; put another way, in neighbourhoods with low median incomes,

vendors are expected to sell more multiple weekly fares than vendors in neighbourhoods with high

median incomes. Moreover, this effect was independent of the percentage of part-time workers in a

neighbourhood. Taken together, these findings suggest that in marginalized neighbourhoods in

Montreal, some low-income earners may purchase multiple weekly fares to compensate for the large

upfront cost of a monthly fare. This finding echoes observational findings in New York (Hickey et al.,

2010). Therefore, our work highlights the need for transit agencies to ensure that their fare schemes are

not unintentionally harming low-income earners who may spend more money over the course of a

month by purchasing multiple weekly fares. Indeed, some transit agencies, such as those in Calgary,

San Francisco, and Seattle, offer subsidies, vouchers or reduced fares to low-income riders (SFMTA,

n.d.; King County Transit, 2015; Calgary Transit, 2015). Since the STM already uses fare smartcards,

the STM could enroll qualifying low-income individuals to be directly eligible for reduced fares. If low-

income individuals cannot afford the upfront price a monthly fare, having a fare-capping mechanism can

permit paying-per-ride without penalizing riders who cannot buy an expensive fare at one time (TfL,

45

n.d.). Future work should survey residents throughout Montreal to enquire into transit fare purchasing

behaviour to directly validate our modelling results.

In the third chapter, I investigated affordability of public transit, which considered not only the

monetary cost of a transit fare, but its relation to the hours of minimum wage work required to purchase

a monthly fare. Indeed, a recent report of the working poor in New York found that the chief concern of

this group was affordability of transit fares, over and above concerns regarding reliability and comfort of

the subway (Stolper & Rankin, 2016). By this affordability indicator alone, compared to other agencies in

large North American cities, a monthly fare in Montreal was the second most affordable, requiring nearly

eight hours of minimum wage work. Moreover, a combined service quality indicator that accounted for

attributes important to riders, society at large, and transit agencies themselves, resulted in the STM

placing third, behind transit agencies in New York and Toronto. Overall, it does seem that more

expensive agencies provide better quality service, and some expensive agencies provide lesser quality

service as well. The methodology provided in Chapter 3 lays the groundwork for researchers and

practitioners wishing to benchmark and compare transit agencies using simple indicators calculated

with publicly reported data, and calls for industry standardization of metrics for reliability and amenities

for customers, like next-arrival information. Nevertheless, future studies can refine the indicators used

here, consider different perspectives, or consider weighting indicators based on rider preference

surveys (Stolper & Rankin, 2016).

A critical reappraisal of transit fare structure is needed in many cities. Raising transit fares for

individuals who can afford to pay more while lowering fares or offering subsidizes for vulnerable

individuals may help address not only equity concerns, but also address funding shortages and improve

transit quality for all (Aggarwala, 2012).

46

REFERENCE LIST

Aggarwala, R. T. (2012). Why Higher Fares Would Be Good for Public Transit. Retrieved 9 April, 2016,

from http://www.citylab.com/commute/2014/07/why-higher-fares-would-be-good-for-public-transit/374314/

APTA. (2007). A Profile of Public Transportation Passenger Demographics and Travel Characteristics Reported in On-Board Surveys. Washington, D.C.: American Public Transportation Association.

APTA. (2016). Public Transport Benefits – Benefits. Retrieved 14 April, 2016, from http://www.apta.com/mediacenter/ptbenefits/Pages/default.aspx

Badami, M. G., & Haider, M. (2007). An analysis of public bus transit performance in Indian cities. Transportation Research Part A: Policy and Practice, 41(10), 961-981.

Calgary Transit. (2015). Low Income Monthly Pass. Retrieved 22 January, 2016, from http://www.calgarytransit.com/fares-passes/passes/low-income-monthly-pass

Carruthers, R., Dick, M., & Saurkar, A. (2005). Affordability of Public Transport in Developing Countries Transport Papers (Vol. 3). Washington, D.C.: World Bank Group.

Cervero, R. (1981). Flat versus differentiated transit pricing: What's a fair fare? Transportation, 10(3), 211-232.

Das, S., & Pandit, D. (2016). Methodology to determine service delivery levels for public transportation. Transportation Planning and Technology, 39(2), 195-217.

de Oña, J., de Oña, R., Eboli, L., & Mazzulla, G. (2013). Perceived service quality in bus transit service: A structural equation approach. Transport Policy, 29, 219-226.

Delbosc, A., & Currie, G. (2011). Exploring the relative influences of transport disadvantage and social exclusion on well-being. Transport Policy, 18(4), 555-562.

Delbosc, A., & Currie, G. (2012). Choice and disadvantage in low-car ownership households. Transport Policy, 23, 8-14.

dell’Olio, L., Ibeas, A., & Cecin, P. (2011). The quality of service desired by public transport users. Transport Policy, 18(1), 217-227.

Diab, E. I., Badami, M. G., & El-Geneidy, A. M. (2015). Bus transit service reliability and improvement strategies: Integrating the perspectives of passengers and transit agencies in North America. Transport Reviews, 35(3), 292-328.

Djurhuus, S., Hansen, H. S., Aadahl, M., & Glümer, C. (2014). The association between access to public transportation and self-reported active commuting. International Journal of Environmental Research and Public Health, 11(12), 12632-12651.

Dolye, A. (2015, 18 December 2015). Minimum Wage Rates for 2014. Retrieved 6 April, 2016, from http://jobsearch.about.com/od/minimumwage/a/minimum-wage-rates-2014.htm

Duggan, P. (2014). Metro wants more money; without it, officials say, rail service might be cut back. Retrieved 14 April, 2016, from https://www.washingtonpost.com/local/trafficandcommuting/metro-wants-more-money-without-it-officials-say-rail-service-might-be-cut-back/2014/11/30/34656f24-7815-11e4-bd1b-03009bd3e984_story.html

Eboli, L., & Mazzulla, G. (2011). A methodology for evaluating transit service quality based on subjective and objective measures from the passenger’s point of view. Transport Policy, 18(1), 172-181.

El-Geneidy, A., Buliung, R., Diab, E., van Lierop, D., Langlois, M., & Legrain, A. (2015). Non-stop equity: Assessing daily intersections between transit accessibility and social disparity across the Greater Toronto and Hamilton Area (GTHA). Environment and Planning B: Planning and Design.

El-Geneidy, A., Grimsrud, M., Wasfi, R., Tétreault, P., & Surprenant-Legault, J. (2014). New evidence on walking distances to transit stops: Identifying redundancies and gaps using variable service areas. Transportation, 41(1), 193-210.

El-Geneidy, A., & Levinson, D. (2006). Access to destinations: Development of accessibility measures. Technical Report MN/RC-2006-16: Minnesota Department of Transportation.

47

El-Geneidy, A., Levinson, D., Diab, E., Boisjoly, G., Verbich, D., & Loong, C. (2016). The cost of equity: Assessing accessibility by transit and social disparity using total travel cost. Paper presented at the 95th Annual Meeting of the Transportation Research Board, Washington, D.C.

Familar, R., Greaves, S., & Ellison, A. (2011). Analysis of speeding behavior: Multilevel modeling approach. Transportation Research Record, 2237, 67-77.

Farber, S., Bartholomew, K., Li, X., Páez, A., & Nurul Habib, K. M. (2014). Assessing social equity in distance based transit fares using a model of travel behavior. Transportation Research Part A: Policy and Practice, 67, 291-303.

Foth, N., Manaugh, K., & El-Geneidy, A. (2013). Towards equitable transit: Examining transit accessibility and social need in Toronto, Canada 1996–2006. Journal of Transport Geography, 29, 1-10.

Geurs, K. T., Boon, W., & Van Wee, B. (2009). Social impacts of transport: Literature review and the state of the practice of transport appraisal in the Netherlands and the United Kingdom. Transport Reviews, 29(1), 69-90.

Geurs, K. T., & van Wee, B. (2004). Accessibility evaluation of land-use and transport strategies: review and research directions. Journal of Transport Geography, 12(2), 127-140.

Giuliano, G. (2005). Low income, public transit, and mobility. Transportation Research Record, 1927, 63-70.

Gkritza, K., Karlaftis, M. G., & Mannering, F. L. (2011). Estimating multimodal transit ridership with a varying fare structure. Transportation Research Part A: Policy and Practice, 45(2), 148-160.

Glaeser, E. L., Kahn, M. E., & Rappaport, J. (2008). Why do the poor live in cities? The role of public transportation. Journal of Urban Economics, 63(1), 1-24.

Gómez-Lobo, A. (2011). Affordability of public transport: A methodological clarification. Journal of Transport Economics and Policy (JTEP), 45(3), 437-456.

Government of Canada (2016, 1 March 2016). Hourly Minimum Wages in CANADA for Adult Workers. Retrieved 12 April, 2016, from http://srv116.services.gc.ca/dimt-wid/sm-mw/rpt2.aspx?lang=eng&dec=5

Grengs, J. (2002). Community-based planning as a source of political change: The transit equity movement of Los Angeles' Bus Riders Union. Journal of the American Planning Association, 68(2), 165-178.

Hickey, R. L. (2005). Impact of transit fare increase on ridership and revenue: Metropolitan Transportation Authority, New York City. Transportation Research Record, 1927, 239-248.

Hickey, R. L., Lu, A., & Reddy, A. (2010). Using quantitative methods in equity and demographic analysis to inform transit fare restructuring decisions. Transportation Research Record, 2144, 80-92.

Hobbs, B. F., & Horn, G. T. F. (1997). Building public confidence in energy planning: A multimethod MCDM approach to demand-side planning at BC gas. Energy Policy, 25(3), 357-375.

Hodge, D. C. (1988). Fiscal equity in urban mass transit systems: A geographic analysis. Annals of the Association of American Geographers, 78(2), 288-306.

Ingram, G. K. (2009). Smart Growth Policies: An Evaluation of Programs and Outcomes. Cambridge, Mass.: Lincoln Institute for Land Policy.

King County Transit. (2015). ORCA LIFT Reduced Fare Program - King County Metro Transit. Retrieved 9 June, 2015, from http://metro.kingcounty.gov/programs-projects/orca-lift/

Krizek, K., & El-Geneidy, A. (2007). Segmenting preferences and habits of transit users and non-users. Journal of Public Transportation, 10(3), 71-94.

Krumholz, N., & Forester, J. (1990). Making Equity Planning Work: Leadership in the Public Sector. Philadelphia: Temple University Press.

Linton, J. (2016). What Factors Are Causing Metro’s Declining Ridership? What Next? – Streetsblog. Retrieved 14 April, 2016, from http://la.streetsblog.org/2016/01/29/what-factors-are-causing-metros-declining-ridership-what-next/

Litman, T. (2002). Evaluating transportation equity. World Transport Policy & Practice, 8(2), 50-65.

48

Manaugh, K., Badami, M. G., & El-Geneidy, A. (2015). Integrating social equity into urban transportation planning: A critical evaluation of equity objectives and measures in transportation plans in North America. Transport Policy, 37, 167-176.

Manaugh, K., & El-Geneidy, A. (2012). Who benefits from new transportation infrastructure? Using accessibility measures to evaluate social equity in public transport provision. In K. Geurs, K. Krizek, & A. Reggiani (Eds.), Accessibility and Transport Planning: Challenges for Europe and North America (pp. 211-227). London, UK: Edward Elgar.

Martens, K. (2012). Justice in transport as justice in accessibility: applying Walzer’s ‘Spheres of Justice’ to the transport sector. Transportation, 39(6), 1035-1053.

Martens, K., Golub, A., & Robinson, G. (2012). A justice-theoretic approach to the distribution of transportation benefits: Implications for transportation planning practice in the United States. Transportation Research Part A: Policy and Practice, 46(4), 684-695.

METRO. (2015). Metro's New Bus Network – Metro business plan & budget FY 2016. Houston. METRO. (2016). Service & Fare Equity Policy. Retrieved 9 April, 2016, from

http://www.ridemetro.org/Pages/FareServiceFareEquityPolicy.aspx MTA. (2015). Subway Ridership Surges 2.6% In One Year. Retrieved 18 March, 2016, from

http://www.mta.info/news-subway-new-york-city-transit/2015/04/20/subway-ridership-surges-26-one-year

Nelson, L. J., & Weikel, D. (2016). Billions spent, but fewer people are using public transportation in Southern California. Retrieved 14 April, 2016, from http://www.latimes.com/local/california/la-me-ridership-slump-20160127-story.html

New York City Transit and Bus Committee. (2016). Transit & Bus Committee Meeting. New York: MTA. Nuworsoo, C., Golub, A., & Deakin, E. (2009). Analyzing equity impacts of transit fare changes: Case study

of Alameda–Contra Costa Transit, California. Evaluation and Program Planning, 32(4), 360-368. Owen, A., & Levinson, D. (2014). Access Across America: Transit 2014. Minneapolis, MN: Center for

Transportation Studies. Pelletier, M.-P., Trépanier, M., & Morency, C. (2011). Smart card data use in public transit: A literature

review. Transportation Research Part C: Emerging Technologies, 19(4), 557-568. Provost, A.-M. (2015, 29 September). Carte OPUS : des tarifs injustes pour les plus pauvres. Retrieved 9

April, 2016, from http://www.journaldemontreal.com/2015/09/29/carte-opus--des-tarifs-injustes-pour-les-plus-pauvres

Randall, E. R., Condry, B. J., & Trompet, M. (2007). Performance measurement development, challenges, and lessons learned. Paper presented at the 86th Annual Meeting of the Transportation Research Board, Washington, D.C.

Ross, N. A., Tremblay, S., & Graham, K. (2004). Neighbourhood influences on health in Montréal, Canada. Social Science & Medicine, 59(7), 1485-1494.

RPA. (2015). Overlooked Boroughs – Where New York City's Transit Falls Short and How To Fix It. New York: Regional Plan Association.

Schuerman, M. (2015). Why MetroCards Are Like Paper Towels. Retrieved 9 June, 2015, from http://www.wnyc.org/story/why-metrocards-are-paper-towels/

Seaman, M., de Cerreno, A., & English-Young, S. (2004). From Rescue to Renaissance: The Achievements of the MTA Capital Program 1982–2004. New York: Rudin Center for Transportation Policy and Management.

Serebrisky, T., Gómez-Lobo, A., Estupiñán, N., & Muñoz-Raskin, R. (2009). Affordability and subsidies in public urban transport: What do we mean, what can be done? Transport Reviews, 29(6), 715-739.

SFMTA. (n.d.). Muni Lifeline Screening Tool. Retrieved 9 June, 2015, from http://www.munilifeline.org/index.cfm

STM. (2015a). OPUS Card. Retrieved 9 June, 2015, from http://www.stm.info/en/info/fares/opus-cards-and-other-fare-media/opus-card

STM. (2015b). Budget 2016. Retrieved 9 June, 2015, from https://www.stm.info/sites/default/files/pdf/fr/budget2016.pdf

49

STM. (2016). Bus Punctuality. Retrieved 12 April, 2016, from http://www.stm.info/en/about/financial_and_corporate_information/all-together-quality-service/bus-punctuality

Stolper, H., & Rankin, N. (2016). The Transit Affordability Crisis: How Reduced MTA Fares Can Help Low-Income New Yorkers Move Ahead. New York: Community Service Society.

Taylor, B. D., & Morris, E. A. (2015). Public transportation objectives and rider demographics: are transit’s priorities poor public policy? Transportation, 42(2), 347-367.

Taylor, B. D., & Norton, A. T. (2009). Paying for transportation: What's a fair price? Journal of Planning Literature, 24(1), 22-36.

Taylor, K. C., & Jones, E. C. (2012). Fair Fare Policies: Pricing Policies That Benefit Transit-Dependent Riders. In M. P. Johnson (Ed.), International Series in Operations Research & Management Science (Vol. 167, pp. 252-272). New York: Springer.

TCRP. (1998). Funding Strategies for Public Transportation (Vol. 1). Washington, D.C. TCRP. (2009). Local and Regional Funding Mechanisms for Public Transportation. Washington, D.C. TCRP. (2013). Transit Capacity and Quality of Service Manual (Third ed.). Washington, D.C.: Transportation

Research Board of the National Academies. TfL. (n.d.). Capping–Transport for London. Retrieved 22 January, 2016, from https://tfl.gov.uk/fares-and-

payments/oyster/using-oyster/price-capping TLS. (2014). Transit Leadership Summit Report 2012–2014. TransitCenter. (2016, 4 April 2016). High Frequency: Why Houston Is Back On The Bus. Retrieved 17

April, 2016, from http://transitcenter.org/2016/04/04/high-frequency-houston-bus-streetfilms/ Trompet, M., Liu, X., & Graham, D. J. (2011). Development of key performance indicator to compare

regularity of service between urban bus operators. Transportation Research Record, 2216, 33-41. TTC. (2016). Daily Customer Service Report. Retrieved 11 April, 2016, from

http://ttc.ca/Customer_Service/Daily_Customer_Service_Report/index.jsp - 1 US DOT. (2013). 2013 Status of the Nation's Highways, Bridges, and Transit: Conditions & Performance.

Washington, D.C. Utsunomiya, M., Attanucci, J., & Wilson, N. (2006). Potential uses of transit smart card registration and

transaction data to improve transit planning. Transportation Research Record, 1971, 119-126. Verbich, D., & El-Geneidy, A. (2016). The pursuit of satisfaction: Variation in satisfaction with bus transit

service among riders with encumbrances and riders with disabilities using a large-scale survey from London, UK. Transport Policy, 47, 64-71.

Wang, Z. J., Li, X. H., & Chen, F. (2015). Impact evaluation of a mass transit fare change on demand and revenue utilizing smart card data. Transportation Research Part A: Policy and Practice, 77, 213-224.

Wasfi, R. A., Ross, N. A., & El-Geneidy, A. M. (2013). Achieving recommended daily physical activity levels through commuting by public transportation: Unpacking individual and contextual influences. Health & Place, 23, 18-25.

Welch, T. F. (2013). Equity in transport: The distribution of transit access and connectivity among affordable housing units. Transport Policy, 30, 283-293.

WMATA. (2015). Metro – Fares. Retrieved 9 June, 2015, from http://www.wmata.com/fares/purchase/passes.cfm